LITERATURE REVIEW

J Neurosurg Pediatr 25:540–547, 2020

Selective dorsal rhizotomy: an illustrated review of

operative techniques
Nebras M. Warsi, MD, CM,1,2 Jignesh Tailor, FRCS(SN), PhD,2 Ian C. Coulter, FRCS(SN), MSc,2
Husain Shakil, MD, MSc,1 Adriana Workewych, BSc,3 Renée Haldenby, MSc(PT),4
Sara Breitbart, MSc,2 Samuel Strantzas, MSc,2 Michael Vandenberk, MSc,2
Michael C. Dewan, MD, MSCI,2 and George M. Ibrahim, MD, PhD1,2
1
Division of Neurosurgery, Department of Surgery, University of Toronto; 2Division of Neurosurgery, Hospital for Sick Children;
3
Faculty of Medicine and 4Department of Pediatrics, University of Toronto, Ontario, Canada

OBJECTIVE Selective dorsal rhizotomy (SDR) is a procedure primarily performed to improve function in a subset of
children with limitations related to spasticity. There is substantial variability in operative techniques among centers and
surgeons. Here, the authors provide a technical review of operative approaches for SDR.
METHODS Ovid MEDLINE, Embase, and PubMed databases were queried in accordance with PRISMA guidelines. All
studies included described a novel surgical technique. The technical nuances of each approach were extracted, includ-
ing extent of exposure, bone removal, and selection of appropriate nerve roots. The operative approach preferred at the
authors’ institution (the “2 × 3 exposure”) is also detailed.
RESULTS Five full-text papers were identified from a total of 380 articles. Operative approaches to SDR varied signifi-
cantly with regard to level of exposure, extent of laminectomy, and identification of nerve roots. The largest exposure
involved a multilevel laminectomy, while the smallest exposure involved a keyhole interlaminar approach. At the Hospital
for Sick Children, the authors utilize a two-level laminoplasty at the level of the conus medullaris. The benefits and disad-
vantages of the spectrum of techniques are discussed, and illustrative figures are provided.
CONCLUSIONS Surgical approaches to SDR vary considerably and are detailed and illustrated in this review as a
guide for neurosurgeons. Future studies should address the long-term impact of these techniques on functional out-
comes and complications such as spinal deformity.
https://thejns.org/doi/abs/10.3171/2019.12.PEDS19629
KEYWORDS selective dorsal rhizotomy; cerebral palsy; spasticity; spine

S
elective dorsal rhizotomy (SDR) is a neurosurgical
treatment primarily for a subset of children with ce-
rebral palsy (CP) characterized by spastic diplegia.
Dorsal rhizotomy was originally practiced in the late 19th
century for the relief of intractable pain by Abbe, Bennett,
and Dana.5 After Sherrington demonstrated that section-
ing dorsal roots could reduce tone in an affected limb,
dorsal rhizotomy was adopted for the treatment of spastic-
ity rather than pain.5 These early procedures carried sub-
stantial morbidity, primarily related to severe sensory loss,
bladder denervation, postoperative weakness, and persis-
tence of spasticity.23
In the 1970s dorsal rhizotomy started to become more
“selective.” In Montpellier, France, Gros introduced intra-
operative electromyography (EMG) to determine the cor-
respondence of each nerve root to specific muscle groups;
those that contributed the most to spasticity were cut,
while rootlets thought to contribute to gait and postural
control were spared.5,23 The procedure also became mul-
tidisciplinary, with physical therapy and electrophysiology
working alongside the surgical team. In parallel, Fasano
et al. described the electrophysiological characteristics
of rootlets thought most likely to contribute to abnormal
muscular tone.6 He advocated for sectioning nerve roots
that had abnormal stimulation thresholds and spread ac-
tivity to muscles not normally innervated by that spinal
level.6 Following his 1987 description of the procedure,
Peacock is credited with popularizing SDR by expanding
upon these previous advances.17
Today, SDR is experiencing a resurgence in popularity

ABBREVIATIONS CP = cerebral palsy; EMG = electromyography; SDR = selective dorsal rhizotomy.

SUBMITTED October 20, 2019. ACCEPTED December 6, 2019.
INCLUDE WHEN CITING Published online February 7, 2020; DOI: 10.3171/2019.12.PEDS19629.

540 J Neurosurg Pediatr Volume 25 • May 2020 ©AANS 2020, except where prohibited by US copyright law

Unauthenticated | Downloaded 05/07/21 11:39 AM UTC


TABLE 1. Summary of operative techniques
Authors & Year Level of Exposure
Peacock et al., 1987 L2–5 Multilevel laminectomy/laminoplasty
Park & Johnston, 2006 Level of conus (determined on
ultrasound)
Bales et al., 2016 Below conus (determined on
MRI)
Sindou & Georgoulis, 2 to 3 levels of interest (most 2- to 3-level interlaminar keyhole Anatomical/EMG based on the position w/in neural
2015 affected nerve roots)
Funk & Haberl, 2016 Level of conus (determined on
MRI)
Current technique (2 × 3 Level of conus (determined on
exposure) MRI)
among neurosurgeons, patients, and healthcare networks.
The rise of social media has strengthened patient and fam-
ily interest in the procedure, while health-economic analy-
ses have demonstrated its cost-effectiveness.2,15 In 2014 the
National Health Service in England commissioned a pro-
spective study across five hospitals to revaluate the state
of funding for the procedure, leading to its recognition
and funding from 2018 onwards.26 However, despite this
growth in popularity, multiple variations in the procedure
have been described with respect to location of exposure,
extent of bone removal, and the identification and selection
of most appropriate roots for sectioning. The present study
aims to serve as a technical review of published methods
for SDR. To this end, we conducted a systematic review of
all described surgical approaches to SDR assessing their
technical nuances, benefits, and potential limitations. We
conclude by describing modifications of the procedure at
Toronto’s Hospital for Sick Children.
Methods
Study Selection
The search protocol was developed according to the
PRISMA guidelines (see Supplementary Fig. 1 for full
search strategy and Supplementary Fig. 2 for PRISMA
checklist).13 Briefly, a literature search was performed on
August 20, 2019, using the keywords “cerebral palsy,” “rhi-
zotomy,” and/or “selective dorsal rhizotomy” in combina-
tion. The databases searched included Ovid MEDLINE,
Embase, and PubMed with no search restrictions. Title,
abstract, and full-text review were conducted by two in-
dependent authors (N.M.W., M.C.D.). Disagreements were
resolved with discussion and re-review. All final studies
included described a novel surgical technique not previ-
ously published.
Data Synthesis
Data pertaining to operative technique—with a particu-
lar emphasis on the extent of exposure, bony removal, and
selection of appropriate nerve roots—were extracted for
each study and are summarized in Table 1. Surgical illus-
trations were created for each approach by a professional
illustrator (A.W.).
Warsi et al.
Extent of Bony Exposure Nerve Root Identification
Simple anatomical/EMG
Single-level laminectomy Anatomical/EMG w/ operating microscope; identify
cleft at conus
Single-level laminectomy EMG thresholds used to separate dorsal & ventral
roots
approach foramen
Single-level laminoplasty Anatomical/EMG w/ operating microscope; cleft at
conus identified
2-level laminoplasty Anatomical/EMG w/ operating microscope; cleft at
conus identified
Results
A total of 380 manuscripts were identified by our initial
search (Supplementary Fig. 1). Overall, 343 studies were
excluded following title and abstract review. Of the 37 full-
text articles that were assessed, 14 reviews were excluded,
11 were nontechnical papers, 3 were repeat descriptions
of previously published techniques, 2 were conference pa-
pers, and 2 were in a foreign language. The remaining 5
articles consisted of original work written in English and
describing a unique operative approach to SDR. These
findings are summarized in Table 1 and detailed graphi-
cally in Fig. 1.
Peacock et al.
In the early 1980s, Peacock and Eastman pioneered
a form of SDR for children with spasticity secondary to
CP, which involved rhizotomies performed in the cauda
equina rather than the conus.18 This technique was later
described in greater detail, coinciding with the reemer-
gence of SDR for the treatment of CP-associated lower-
limb spasticity in the modern era.17
Under general anesthesia, patients are positioned prone
and a midline incision is made to expose the lumbar ver-
tebrae. Partial L2–5 laminectomies are undertaken with
adjacent facet joint preservation. Following a longitudi-
nal dural opening and arachnoid dissection, the posterior
roots of L2–S1 are identified and isolated. Utilizing two
unipolar probes, rootlets of each posterior root are sequen-
tially subjected to electrical stimulation via single and te-
tanic stimuli in the range of 10–150 V, as described by
Fasano et al.6,7 The corresponding muscle responses were
observed both visually and with EMG.
Criteria that support the decision to divide the rootlet/s
of interest include a low threshold associated with a sus-
tained muscular contraction or spread of contraction to
contiguous muscle groups beyond the rootlet’s segmental
distribution. Conversely, rootlets are left intact if thresh-
olds are relatively high and muscle contraction is brief
without diffusion.
Park and Johnston
In 2006, Park and Johnston16 subsequently reported
J Neurosurg Pediatr Volume 25 • May 2020 541
Unauthenticated | Downloaded 05/07/21 11:39 AM UTC
Warsi et al.

FIG. 1. Illustrations of the different bony exposures in SDR. Peacock and colleagues’ 1987 technique17 utilized a multilevel
laminectomy below the conus; in 2006, Park and Johnston16 modified this to a single-level laminectomy at the conus; Sindou and
Georgoulis22 used a less-invasive but more challenging keyhole interlaminar exposure; in 2016, Funk and Haberl8 described a
single-level laminoplasty at the conus; that same year, Bales et al. modified Park’s exposure to a single-level laminectomy below
the conus, with EMG to identify relevant nerve roots; and finally, our so-called 2 × 3 approach is based on a two-level laminoplasty
at the level of the conus. Copyright Nebras Warsi. Published with permission.

542 J Neurosurg Pediatr Volume 25 • May 2020

Unauthenticated | Downloaded 05/07/21 11:39 AM UTC


FIG. 2. Ultrasound and MR images of the CSF “cleft” used to distinguish dorsal and ventral roots. A: Sagittal ultrasound view
of the conus medullaris below bony exposure. B: Axial ultrasound view at the conus showing the CSF cleft used to distinguish
dorsal and ventral roots (arrowheads [upper]) and below the conus (lower), where the cleft can no longer be appreciated. C: MRI
correlation of the CSF cleft as seen on ultrasonic views (upper image at the conus and lower image below). Figure is available in
color online only.
their experience of limiting the lumbar exposure.16 Follow-
ing induction of general anesthesia, patients are positioned
prone on gel bolsters in Trendelenburg angulation to en-
courage rostral pooling of cerebrospinal fluid (CSF). In
children younger than 10 years of age, ultrasound images
of the conus and cauda equina are obtained to guide the
approach. In older children, where this would not be pos-
sible, the L1 spinous process is localized using x-ray imag-
ing and marked with infiltration of indigo carmine dye.
Following a limited midline lumbar incision over the
interspinous space overlying the conus, the interspinous
ligament and ligamentum flavum are excised to permit
further ultrasonic views. Typically, two levels are exam-
ined to localize the conus and cauda equina. A laminecto-
my over the index level is performed using a craniotome to
minimize facet joint disruption, whereupon the ultrasound
is reemployed to reassess the conus’ position to determine
whether additional caudal laminectomies are necessary to
expose at least 5 mm of the caudal conus. This allows safe
separation of the dorsal and ventral roots.
A linear dural incision is made and the operating mi-
croscope is then introduced. The operating table may be
tilted away from the surgeon to permit dissection of the
contralateral nerve roots. The arachnoid is divided and the
conus and filum terminale are visualized. The L2 nerve
roots are next identified at their respective foramina. The
dorsal and ventral roots are separated and traced back
to the conus in order to appreciate the cleft that defines
ventral and dorsal sections (Fig. 2). Following medial dis-
placement of the L2 root and neighboring dorsal roots, the
ventral roots are covered with a cottonoid. Next the sur-
geon is advised to identify the S2 root and lower sacral
Warsi et al.
roots (S3–5). The lower sacral nerves characteristically
have little space between them and are left intact. A Silas-
tic sheet is positioned under the dorsal L2–S2 nerve roots
to gently separate them from the ventral roots as the pro-
cedure proceeds.
Once the dorsal roots are spread on the Silastic sheet,
EMG testing on each root takes place to establish an in-
nervation pattern using the hooks of the rhizotomy probes.
Threshold voltages are determined via recordings from
the lower-limb musculature. The root is then dissected
into 3–5 rootlets or fascicles of equal size using a Scheer
needle and stimulated with constant square-wave pulses
of 0.1 msec at a rate of 0.5 Hz. An increasing stimulus
is applied until a reflex response is visualized from the
ipsilateral muscles. When a reflex threshold is determined,
the rootlet is then subjected to a train of a 50-Hz tetanic
stimulation for 1 second and the resulting reflex graded.
Park and Johnston16 based their decision to divide a giv-
en rootlet on the number of rootlets producing sustained
responses at that level and their response intensity, as de-
scribed by Phillips and Park.19 Rootlets producing a signif-
icant reflex response (typically 3+ or 4+) undergo division.
At least one rootlet is left intact in order to preserve sensa-
tion. Deafferented roots are sequentially placed below the
Silastic sheet to separate them from the untested bundle.
The authors report sectioning 60%–65% of the examined
nerve roots. Half of the L1 root is routinely divided with-
out EMG testing as it was deemed unreliable for this level.
The dura mater is closed with a 4-0 monofilament nylon.
An intradural injection of clonidine and morphine is ap-
plied, and Gelfoam is left over the laminectomy defect
before a layered wound closure is performed. Since their
J Neurosurg Pediatr Volume 25 • May 2020 543
Unauthenticated | Downloaded 05/07/21 11:39 AM UTC
Warsi et al.
initial publication, the group has also described the addi-
tion of a postoperative epidural catheter to improve patient
analgesia.14
Bales et al.
In 2016, Bales et al.1 described a modification of Park’s
technique that selectively analyzed each individual nerve
root with EMG to separate dorsal and ventral nerve roots
through comparison of stimulus responses. They argued
that the anatomical variation present at the level of the
conus can make Park’s anatomical approach to identify-
ing the dorsal roots particularly challenging, and therefore
they advocated for the use of electrophysiology to dichoto-
mize the rootlets.
Preoperative MRI is performed to determine the level
of the conus. After induction of general anesthesia, the pa-
tient is placed in the prone position and the laminectomy
is planned for one vertebral level below the conus, typi-
cally at L2. An intraoperative radiograph is taken to verify
the location of the appropriate level, which is then marked
at the level of the skin. A single-level laminectomy is per-
formed, and intraoperative ultrasound is used to visualize
the nerve roots and ensure the exposure is below the level
of the conus. Care is taken to remain medial to the facets
to minimize the risk of postlaminectomy spinal deformity.
The dura and arachnoid are opened, and a Silastic sheet
cut to a width of 3–4 mm is placed ventrally around all the
roots of the cauda equina.
Direct stimulation of the nerve rootlets is achieved with
bipolar stimulation using rhizotomy probes. The initial
goal is to differentiate the ventral and dorsal nerve root-
lets by recording motor thresholds. Motor responses fol-
lowing direct stimulation of ventral (motor) rootlets are
typically achieved at a range of less than 0.4–0.5 mA and
most commonly stimulate at 0.1 mA. By comparison, dor-
sal (sensory) rootlets have thresholds greater than 0.5 mA
and most commonly stimulate above 1.0 mA. Rootlets that
stimulate at low thresholds (typically < 0.4 mA) are ex-
cluded as ventral (motor) rootlets and are tucked behind
the Silastic dam. Rootlets that elicited an anal sphincter
response are also preserved. Rootlets that stimulate at the
established threshold for sensory rootlets are further stim-
ulated with a 1-second train of 50-Hz tetanic stimulation.
Dorsal rootlet responses are graded based on Park’s
criteria, as previously described.16 If the response is mark-
edly abnormal, 75%–90% of the rootlet is cut. If slightly
abnormal, 50% of the rootlet is cut. If the response to
tetanic stimulation is normal but the rootlet elicits a re-
sponse only in affected muscle groups, 50% of the rootlet
is incised. Once all the rootlets have been stimulated in
this fashion, the number of rootlets that are incised is re-
viewed to ensure that there is adequate representation of
the affected muscle groups.
Sindou et al.
In 2015, Sindou and Georgoulis22 described the keyhole
interlaminar dorsal rhizotomy technique in an attempt to
overcome the potential disadvantages of multilevel lami-
nectomy. The goals were to achieve anatomical accuracy
of the ventral and dorsal nerve roots, functional selectivity
544 J Neurosurg Pediatr Volume 25 • May 2020
through electrophysiology, and reduction of the risks of
spinal instability.
Anatomical accuracy is achieved by surgically access-
ing roots where both their ventral and dorsal components
are clearly separated at the exit to/entry from their forami-
nal sheath. This anatomical localization is then followed
by functional verification of the dorsal root by electrical
stimulation and subsequent interruption of the dorsal root/
rootlets likely to participate in spasticity. According to
the authors, the risk of spinal instability is minimized by
maximal preservation of the bony and posterior ligamen-
tous structures through an interlaminar spinal approach.
Preoperatively, afferents from the most affected mus-
cles are identified based on anatomical knowledge. These
specific dorsal nerve roots are then targeted through 2 or 3
interlaminar spaces. For example, the transiting L3 and L4
dorsal roots could be accessed through an L2–3 interlami-
nar keyhole. The patient is positioned prone, and a midline
lumbar incision is performed according to the topography
of the predetermined interlaminar spaces. An intraopera-
tive radiograph is used to confirm the operative levels. The
muscles are dissected, while leaving the spinous processes
and interspinous ligaments intact. At the selected inter-
laminar spaces, the ligamentum flavum is removed and
the space is expanded with rongeurs. The lower two-thirds
of the upper lamina and the upper two-thirds of the lower
lamina are removed in the midline underneath the inter-
spinous ligaments.
A 2-cm midline opening is made in the dura and arach-
noid. The microscope is then employed, and an oblique
trajectory, on the order of 45°, is used to access the con-
tralateral roots by passing underneath the preserved in-
terspinous ligaments. Through this approach, Sindou et
al.,23 have found that the ventral and dorsal components of
the nerve root are identified constantly at the exit to/entry
from their corresponding dural sheath. Two roots can be
accessed on each side, one upward and one downward. At
this point, the posteriorly located dorsal root with its 4–5
rootlets are tested with electrophysiology, as previously
described, and subsequently sectioned if selected based
on the Phillips and Park criteria.19
Funk and Haberl
In 2016, Funk and Haberl8 published a paper describing
a single-level laminoplasty approach. Their aim was again
to minimize instability of the spine, and by extension re-
duce the incidence of late scoliosis.
Their technique begins with preoperative MRI to lo-
cate the level of the conus medullaris. The patient is
placed prone, and a laminectomy utilizing a craniotome
drill at the level of the conus is used to provide interlami-
nar access. The authors then describe the use of long-blade
Ruskin bone-splitting forceps to make oblique cuts in the
spinous processes above and below the level where the
laminae are dissected. With these cuts, a midline complex
is disconnected from the rest of the spine and stored in a
damp cloth. This complex consists of the spinous process
of the MRI-defined level, the adjoining medial lamina, the
adjacent superior and inferior spinous process segments
attached through the intervening posterior ligaments.
A midline dural incision follows, and the L1–S2 dorsal
Unauthenticated | Downloaded 05/07/21 11:39 AM UTC

sensory roots are separated and placed in a Silastic sheet.
For the L1 rootlet, 50% of the sensory rootlets are cut
without EMG. From L2 to S1, at each given level, roots are
separated into anywhere from 4 to 8 rootlets. According to
EMG evaluation using the Phillips and Park grading scale,
50%–60% of the rootlets are cut.19 The S2 root is divided
into 2 parts, the bulbocavernous reflex is stimulated, and
the fascicle showing the least pudendal nerve activity is
severed.
Following the SDR, the dura is closed tightly, and the
aforementioned midline complex is replated to the spine,
across the intact spinous process and dissected lamina.
This is done using multihole miniplates. The superior and
inferior obliquely cut spinous processes are fixed to their
respective ends using single sutures. In younger children,
this suturing is done once through the bony spinous pro-
cess. In older children, the stitch is placed twice around the
spinous process fragments through the interspinous liga-
ments. Muscle and fascia are closed with sutures running
through the interspinous and supraspinous ligaments.
The 2 × 3 Exposure Technique
Patients are initially assessed in a multidisciplinary
clinic. Candidates are typically 4–8 years old, with am-
bulatory spastic diplegic CP. Clinical examinations and
EMG recordings from the L1–S1 muscle groups are used
to form a preoperative rhizotomy plan. This plan is made
in consensus with the neurosurgical and physiotherapy
teams. The anticipated amount of dorsal root transection
at each level is generally on the order of 50%–70%.
A preoperative MR image of the spine is used to identi-
fy the level of the conus so as to inform the ideal location of
a 2-level laminoplasty aimed at unroofing the L2–S1 nerve
roots. With the patient in the prone position, intraoperative
fluoroscopy is used to identify the target levels for the lami-
noplasty, and these levels are demarcated on the skin. A
clear drape is placed inferiorly and tented over a curtain to
allow physiotherapists access to the lower extremities.
A midline dissection is performed until the supraspi-
nous ligament overlying the planned laminoplasty site
is identified. Bilateral subperiosteal dissections are then
performed while sparing this ligament. A 2-level lamino-
plasty respecting the medial facet is then performed (Fig.
1). Inferiorly, the supraspinous ligament, interspinous liga-
ment, and ligamentum flavum are divided. The spinous
processes and posterior ligament complex above the target
site are preserved and reflected superiorly. The ultrasound
probe is used to confirm the level of the conus, as well as
to appreciate the cleft between dorsal and ventral roots vis-
ible only at this level (Fig. 2). This allows us to identify the
optimal location for the midline dural opening. The dural
leaflets are then tacked dorsolaterally to fascia to maxi-
mize intradural visualization.
The method is described herein as a 2 × 3 exposure
because two nerve roots (most commonly L2 and L3) can
be visualized at the dural exit site, and 3 nerve roots (most
commonly L4, L5, and S1) are identified in the midline. L2
is readily identified beneath the L2 pedicle, and anatomi-
cally, the dorsal and ventral roots are separated and L2 is
sectioned. A Silastic sheet is then placed in the CSF cleft
at the conus separating the dorsal and ventral roots from
Warsi et al.
FIG. 3. Schematic representation of rootlet division. As shown, a Silastic
sheet is placed between the dorsal and ventral nerve roots. The dorsal
(sensory) nerve roots are then divided into individual rootlets using
microdissectors. Stimulation of each individual rootlet then proceeds as
described. Copyright Nebras Warsi. Published with permission.
L3 to S1 (Fig. 3). The L3 nerve root is followed inferiorly
toward its lateral exit site in the dura, which can be visu-
alized by sharply angling the operating microscope. This
provides final anatomical confirmation of the separation of
motor and sensory roots. From lateral to medial, L4 to S1
are stimulated and sectioned systematically thereafter. We
repeat the procedure on the contralateral side, and finally,
when indicated, L1 may be sectioned by sharply angling
the microscope superiorly. If indicated, the exposure also
provides opportunity to selectively section the S2 dorsal
rootlets, as previously described, although this is not rou-
tinely performed for all patients.
Our stimulation protocol uses a bipolar probe with a
100-μsec pulse width, which is delivered at 2.1 Hz to the
left and right L2–S1 nerve roots in order to obtain both
sensory and motor stimulation intensity thresholds. At
each level, both dorsal sensory and ventral motor roots
are identified and separated with a Silastic sheet. Rootlets
within the sensory roots are dissected apart using micro-
dissectors. The S2–4 nerve roots are confirmed via stimu-
lation and are spared.
Tetanic stimulation at sensory thresholds is then ap-
plied to separate rootlets at a frequency of 50 Hz for 1 sec-
ond. We grade responses to stimulation of rootlets using
the Phillips and Park classification system.19 Rootlets with
the most abnormal responses are transected. The number
of rootlets transected at a given level is based on the pre-
operative rhizotomy plan described above. This process is
continued systematically to bilateral L2–S1 nerve roots.
Evoked potentials are then used to confirm the functional
integrity of the ventral roots prior to closure.
The dura is closed in a watertight fashion, and the lami-
noplasty complex is reflected back into position and fixed
to native bone using miniplates. The remaining closure
process proceeds in standard multilayer fashion for the
lumbar spine.
Discussion
As interest in SDR continues to grow, there is a need to
J Neurosurg Pediatr Volume 25 • May 2020 545
Unauthenticated | Downloaded 05/07/21 11:39 AM UTC
Warsi et al.
evaluate and compare the various surgical nuances of this
procedure. A variety of techniques have been described,
which vary with regard to the extent of bony exposure,
identification of dorsal and ventral nerve roots, and selec-
tion of appropriate rootlets for sectioning (Table 1).
Peacock et al.17 have described the largest bony expo-
sure, which may be associated with an increased incidence
of spinal deformity.10 Alternatively, 1- and 2-level proce-
dures come at the cost of easy anatomical identification
of individual nerve roots. To circumvent this problem, the
use of extended EMG recordings increases the specificity
of nerve root identification.11 The technique described by
Bales et al.1 has considerable overlap with that initially de-
scribed by Park and Johnston.16 However, a key difference
is that Bales et al. make their opening below the conus
and identify nerve roots using EMG, which is an entirely
different process from the anatomical identification de-
scribed by Park and Johnston. Sindou and colleagues22,23
described a keyhole interlaminar approach, whereas Funk
and Haberl used a single-level laminoplasty. Our 2-level
laminoplasty potentially mitigates the risk of multilevel
laminectomy with several advantages over a single-level
approach. Cranially, the anatomy of the conus, and the
CSF cleft between dorsal and ventral roots is preserved
(Fig. 2), facilitating separation of the ventral and dorsal
roots. Caudally, two levels of exiting nerves (typically
L2 and L3) can be identified at the dural exit site, pro-
viding confirmation of the separation between motor and
sensory roots. Distal to the conus, the sensory roots are
best appreciated and can be traced backward toward the
conus. Furthermore, with regard to bony exposure, there
is evidence to suggest that laminoplasty confers increased
biomechanical stability to the spine compared with lami-
nectomy, although the exact extent of this advantage is
unclear.10,25
With regard to postoperative spinal deformity, there is a
growing body of literature to support the long-term risks of
this complication with multilevel bony exposures.3,4,9,20,24,27
The incidence of postoperative spinal deformity ranges
from approximately 12% to greater than 35% in the lit-
erature.3,4,9,20,24,27 Interestingly, even without surgery Saito
et al. estimated the rate of spinal deformity in CP to be
68%, especially scoliosis, with large multilevel exposures
thought to increase this risk.21 Unfortunately, at present,
no comparative studies exist that have assessed whether
approaches such as the keyhole interlaminar approach and
laminotomy techniques offer a significant advantage in
this regard.
Identification of appropriate nerve roots is another chal-
lenge in SDR as the procedure by which candidate nerve
roots are targeted can vary widely based on the operative
approach. For example, many of the procedures described
require tracing back each root with an operative micro-
scope and identifying the cleft at the conus to separate
dorsal and ventral roots. The benefits of these conus-level
approaches are that they allow a more limited bony expo-
sure. In our experience and in the literature, this procedure
has been successful but does require thorough knowledge
of the anatomy at the level of the conus. Similarly, for the
keyhole interlaminar approach, intimate operative experi-
ence with lumbar laminoforaminotomy techniques is criti-
546 J Neurosurg Pediatr Volume 25 • May 2020
cal for adequate exposure and identification of relevant
nerve roots. In this regard, Peacock and colleagues’ initial
approach is simplest,17 followed by Bales et al.’s dichoto-
mizations of nerve roots using EMG.1
Beyond identification of nerve roots, the extent to which
each root should be transected is unclear. There is evi-
dence to suggest a direct relationship between the percent-
age of dorsal root transected and postoperative functional
improvement.12 At our institution, we employ preoperative
assessments to determine the amount to transect at each
level; however, this is not widely accepted. Intraopera-
tively, we only transect rootlets with the most abnormal
responses, and our target is on the order of 50%–70% per
level. This is in keeping with the technique described by
Funk and Haberl. However, it is not clear whether it is
more important to transect simply a larger percentage of
the dorsal root or simply a larger percentage of the subset
of abnormal rootlets.
Given that no single approach is clearly superior, the
technique used will depend on individual surgeon prefer-
ence and comfort with the variable interrelationship be-
tween anatomy and neurophysiology in this patient popu-
lation.
Conclusions
A number of questions remain to be answered as they
pertain to these different operative approaches to SDR.
One critical avenue of further study will be a comparison
of each of these operative techniques vis-à-vis the out-
comes and potential complications. Similarly, it would be
important to evaluate these factors among different centers
and surgeons, as significant heterogeneity has been previ-
ously reported. For example, Park’s group, reported only a
single CSF leak requiring operative repair in their series of
1500 patients and no spinal deformity requiring operation,
while other articles put the incidence of significant spinal
deformity in the 10% range.3,4,9,16,20,24,27 The value of ad-
juncts, including intraoperative neuromonitoring, should
also be the subject of future research.
References
1. Bales J, Apkon S, Osorio M, Kinney G, Robison RA, Hooper
E, et al: Infra-conus single-level laminectomy for selective
dorsal rhizotomy: technical advance. Pediatr Neurosurg
51:284–291, 2016
2. Canty MJ, Breitbart S, Siegel L, Fehlings D, Milo-Manson
G, Alotaibi NM, et al: The role of social media in selective
dorsal rhizotomy for children: information sharing and social
support. Childs Nerv Syst 35:2179–2185, 2019
3. Cobb MA, Boop FA: Replacement laminoplasty in selective
dorsal rhizotomy: possible protection against the develop-
ment of musculoskeletal pain. Pediatr Neurosurg 21:237–
242, 1994
4. Crawford K, Karol LA, Herring JA: Severe lumbar lordosis
after dorsal rhizotomy. J Pediatr Orthop 16:336–339, 1996
5. Enslin JMN, Langerak NG, Fieggen AG: The evolution of
selective dorsal rhizotomy for the management of spasticity.
Neurotherapeutics 16:3–8, 2019
6. Fasano VA, Broggi G, Zeme S: Intraoperative electrical stim-
ulation for functional posterior rhizotomy. Scand J Rehabil
Med Suppl 17:149–154, 1988
7. Fasano VA, Broggi G, Zeme S, Lo Russo G, Sguazzi A:
Unauthenticated | Downloaded 05/07/21 11:39 AM UTC

Long-term results of posterior functional rhizotomy. Acta
Neurochir Suppl (Wien) 30:435–439, 1980
8. Funk JF, Haberl H: Monosegmental laminoplasty for selec-
tive dorsal rhizotomy—operative technique and influence
on the development of scoliosis in ambulatory children with
cerebral palsy. Childs Nerv Syst 32:819–825, 2016
9. Golan JD, Hall JA, O’Gorman G, Poulin C, Benaroch TE,
Cantin MA, et al: Spinal deformities following selective dor-
sal rhizotomy. J Neurosurg 106 (6 Suppl):441–449, 2007
10. Johnson MB, Goldstein L, Thomas SS, Piatt J, Aiona M,
Sussman M: Spinal deformity after selective dorsal rhi-
zotomy in ambulatory patients with cerebral palsy. J Pediatr
Orthop 24:529–536, 2004
11. Legatt AD, Schroeder CE, Gill B, Goodrich JT: Electrical
stimulation and multichannel EMG recording for identifica-
tion of functional neural tissue during cauda equina surgery.
Childs Nerv Syst 8:185–189, 1992
12. McLaughlin J, Bjornson K, Temkin N, Steinbok P, Wright V,
Reiner A, et al: Selective dorsal rhizotomy: meta-analysis of
three randomized controlled trials. Dev Med Child Neurol
44:17–25, 2002
13. Moher D, Liberati A, Tetzlaff J, Altman DG: Preferred re-
porting items for systematic reviews and meta-analyses: the
PRISMA statement. PLoS Med 6:e1000097, 2009
14. Moore RP, Wester T, Sunder R, Schrock C, Park TS: Peri-
operative pain management in children with cerebral palsy:
comparative efficacy of epidural vs systemic analgesia proto-
cols. Paediatr Anaesth 23:720–725, 2013
15. National Institute for Health and Care Excellence: Final
Report: Selective Dorsal Rhizotomy (SDR). London:
Centre for Health Technology Evaluation Observational Data
Unit, 2018 (https://www.england.nhs.uk/commissioning/
wp-content/uploads/sites/12/2019/04/SDR-CtE-NICE-
Evaluation-Report.pdf) [Accessed December 20, 2019]
16. Park TS, Johnston JM: Surgical techniques of selective dorsal
rhizotomy for spastic cerebral palsy. Technical note. Neuro-
surg Focus 21(2):e7, 2006
17. Peacock WJ, Arens LJ, Berman B: Cerebral palsy spasticity.
Selective posterior rhizotomy. Pediatr Neurosci 13:61–66,
1987
18. Peacock WJ, Eastman RW: The neurosurgical management
of spasticity. S Afr Med J 60:849–850, 1981
19. Phillips LH, Park TS: Electrophysiology studies of selective
posterior rhizotomy patients, in Park TS, Phillips LH, Pea-
cock WJ (eds): Management of Spasticity in Cerebral Pal-
sy and Spinal Cord Injury. Neurosurgery. State of the Art
Reviews. Philadelphia: Hanley & Belfus, 1989, pp 459–469
20. Ravindra VM, Christensen MT, Onwuzulike K, Smith JT,
Halvorson K, Brockmeyer DL, et al: Risk factors for progres-
sive neuromuscular scoliosis requiring posterior spinal fu-
sion after selective dorsal rhizotomy. J Neurosurg Pediatr
20:456–463, 2017
21. Saito N, Ebara S, Ohotsuka K, Kumeta H, Takaoka K:
Natural history of scoliosis in spastic cerebral palsy. Lancet
351:1687–1692, 1998
Warsi et al.
22. Sindou M, Georgoulis G: Keyhole interlaminar dorsal rhizot-
omy for spastic diplegia in cerebral palsy. Acta Neurochir
(Wien) 157:1187–1196, 2015
23. Sindou M, Georgoulis G, Mertens P: History of neurosurgi-
cal treatment for spasticity, in Neurosurgery for Spasticity.
A Practical Guide for Treating Children and Adults. Vi-
enna: Springer, 2014
24. Steinbok P, Hicdonmez T, Sawatzky B, Beauchamp R,
Wickenheiser D: Spinal deformities after selective dorsal
rhizotomy for spastic cerebral palsy. J Neurosurg 102 (4
Suppl):363–373, 2005
25. Subramaniam V, Chamberlain RH, Theodore N, Baek S,
Safavi-Abbasi S, Senoğlu M, et al: Biomechanical effects
of laminoplasty versus laminectomy: stenosis and stability.
Spine (Phila Pa 1976) 34:E573–E578, 2009
26. Summers J, Coker B, Eddy S, Elstad M, Bunce C, Bourmpaki
E, et al: Selective dorsal rhizotomy in ambulant children with
cerebral palsy: an observational cohort study. Lancet Child
Adolesc Health 3:455–462, 2019
27. Turi M, Kalen V: The risk of spinal deformity after selective
dorsal rhizotomy. J Pediatr Orthop 20:104–107, 2000
Disclosures
The authors report no conflict of interest concerning the materi-
als or methods used in this study or the findings specified in this
paper.
Author Contributions
Conception and design: Ibrahim, Warsi, Dewan. Acquisition of
data: Warsi, Tailor, Coulter, Shakil. Analysis and interpretation
of data: Ibrahim, Warsi, Tailor, Coulter, Shakil, Breitbart, Dewan.
Drafting the article: Warsi, Tailor, Coulter, Shakil, Workewych,
Haldenby. Critically revising the article: all authors. Reviewed
submitted version of manuscript: all authors. Approved the final
version of the manuscript on behalf of all authors: Ibrahim.
Statistical analysis: Administrative/technical/material support:
Haldenby, Breitbart, Strantzas, Vandenberk. Study supervision:
Ibrahim, Dewan. Illustrations: Workewych.
Supplemental Information
Online-Only Content
Supplemental material is available with the online version of the
article.
Supplementary Figs. 1 & 2. https://thejns.org/doi/suppl/​
10.3171/​2019.12.PEDS19629.
Correspondence
George M. Ibrahim: Hospital for Sick Children, Toronto, ON,
Canada. george.ibrahim@sickkids.ca.
J Neurosurg Pediatr Volume 25 • May 2020 547
Unauthenticated | Downloaded 05/07/21 11:39 AM UTC